Analysis of Black Lives Matter

Analysis of Black Lives Matter

On February 26 2012, George Zimmerman, a neighborhood ‘watchperson’ shot and killed 17-year old black teenager Trayvon Martin. More than one year later, in July 2013, the hashtag #BlackLivesMatter was used for the first time in a Facebook post about the acquittal of George Zimmerman. Now, 5 years after the fatal shooting, this hashtag has been used more than 740874 times by 311525 different twitter users.

Written by Ada Pozo
Figure 1. Comparison of the number of tweets in #BlackLivesMatter and #AllLiveMatters aggregated by day.

BlackLivesMatter movement against police brutality and racism in US consolidated hashtag activism as a new way of protesting in social media. As a response to it, other hashtags supporting the police have appeared: #AllLivesMatter, #BlueLivesMatter, #WhiteLivesMattter, #NYPDLivesMatter and #PoliceLivesMatter. These hashtags have been used 387215 times (almost half less than #BlackLivesMatter) by 166281 users. The most widespread of these hashtags, #AllLivesMatter, was created before this protest, in 2012, and adopted by the people against it.

The popularity of the hashtags on both sides of the protest, which we’ll call #BlackLivesMatter and #AllLivesMatter, grew slowly. As can be observed in Figure 1, which shows the number of tweets of each protest over time, the popularity and importance of the movement did not rise until the summer of 2014. It was prompted by the death, on July 17, of Eric Gardner, who was killed with a chokehold by a police officer in Long Island, New York, prompting the appearance of the hashtag #ICantBreathe; and, especially by the shooting of Michael Brown the 9th of August in Ferguson, Missouri by a white policeman, which resulted in the #Ferguson hashtag.

Despite this, the biggest rise in the use of #BlackLivesMatter occurred in November 2014. On the 22nd 12-year-old Tamir Rice was killed by a police officer while playing with a toy gun in a playground. Two days later, on the 24th, a grand jury decided not to indict the police officer who killed Michael Brown. Protest ensued, combined with a rise on the number of people tweeting the #BlackLivesMatter hashtag and also an increase of the use, though much smaller, of #AllLivesMatter.

The next rise can be found on the 3rd December, when the no indictment in the Gardner case was announced. This resulted again in new protests, but the event that had the biggest impact took place on the 9th of December, when LeBron James, Kobe Bryant and other NBA players wore an #ICantBreathe t-shirt during a game, referring to the Eric Gardner case. In the following days, there were more than 62121 tweets from #BlackLivesMatter. This event also supposed a big rise in #AllLivesMatter tweets, with a significant number of tweets whereas before there were very few. The protests and the agitation continued during the following days and on December 20, two police officers were killed in New York while they were sitting in their patrol car. After these events, even though the number of tweets decremented, both hashtags appeared regularly, with rises due to concrete events, such as the one-year anniversary of the death of Michael Brown; the death of Sandra Bland in police custody, three days after being pulled-over; or the defense of the #BlackLivesMatter protest by Bernie Sanders in the presidential campaign. The most important rise since then, however, was not due to a unique event, but to the deaths of 36 black people in the hands of the police just in March 2015.

Overall, Figure 1 shows that the growth of the use of the #BlackLivesMatter and #AllLivesMatter hasthags has been parallel, but #BlackLivesMatter has a much bigger support, followers and success. In addition, the impact of different events is clear, with rises on the number of tweets on the days following the protests. However, this rise is clearly dependent on the type of event. In three occasions, after the death of Tamir Rice and the no indictments in both the Brown case and the Gardner case, we can see a rise to around 1600 and 100 tweets with #BlackLivesMatter and #AllLivesMatter hashtags, respectively. Nevertheless, the event that marked the bigger number #BlackLivesMatter’s tweets, with 62121 in the following days, and a regular use of #AllLivesMatter, was the NBA match.

The biggest number of tweets for #AllLivesMatter in a day, 15400, was caused by the death of the two police officers in New York. This event did not provoke such a big reaction of the #BlackLivesMatter hashtag, with only 4360 tweets. It is interesting to notice, as well, that the number of tweets after an event related to #AllLivesMatter tends to decay much more rapidly than in those related #BlackLivesMatter, where the rise lasts for several days.

Table 1. Numbers on the impact of both sides of the protest.
  #BlackLivesMatter #AllLivesMatter
Number of tweets 846415 387206
Number of users 342192 166280
More frequent languages English, German, Spanish, French, Italian and Dutch English, German, Spanish, French, Italian and Dutch
Percentage in English 97% 96%

Are the hashtags polarized towards just one side of the protest?

o far, we have assumed that both sides of the protest use different hashtags and that these hashtags are really polarized towards supporting #BlackLivesMatter or the police. Now, we analyze whether this is true by showing which percentage of tweets contained hashtags in support of the police in addition to #BlackLivesMatter aggregated by month.

Figure 2. Occurence of hashtags in favour and against the protest together aggregated by month.

Figure 2 shows that only the hashtags #AllLiveMatters, #NYPDLivesMatter and #PoliceLivesMatter were tweeted together with #BlackLivesMatter. The percentage of tweets is non-significant and indicates a great polarization except for #AllLivesMatter, which seems to be a little bit less polarized. It is also important that this polarization grows with time, with a much higher percentage of appearance with #BlackLivesMatter in the first month that drops later.


Whats discussed in the tweets?

Figure 3. Topics in #BlackLivesMatter tweets. A smaller value of relevance on the right slider selects the keywords most repeated for this topic whose overall frequency in all tweets is smaller.

The tweets in support of #BlackLivesMatter were focused in four ideas in general:

  • Crying out about police brutality and the numerous deaths of unarmed black people .
  • Demanding justice and criticizing the judicial system.
  • Centered in concrete events like the death of Michal Brown, Eric Gardner or Tamir Rice.
  • Discussing about race in general.

The topics found, usually, use one or two of these ideas and are very closely related to one another. It is interesting to observe how, despite seeming to be a triggering event for the number of tweets, the NBA game does not seem to have had an importance or to have been discussed.

Figure 4. Topics in #AllLivesMatter tweets. A smaller value of relevance on the right slider selects the keywords most repeated for this topic whose overall frequency in all tweets is smaller.

On the other hand, the tweets against the protest cover much more heterogeneous topics, with the primary idea being the defense of the police and that their actions were incited and justified. Among the most prominent topics are also racist messages, with keywords like #killallmuslims or ‘terrorist’. However, it should also be noted that the idea of reducing the violence and stopping the hate appears in many of the tweets . Lastly, #AllLivesMatter seems to be also used to defend other causes, like transsexual, abortion and anti-abortion, a reference of this hashtag not being so polarized.

Table 2. Most repeated words among supporters and detractors of the protest.
  Most used keywords
#BlackLivesMatter #BlackLivesMatter, black, police, #icantbreathe, people, #ferguson, justice, #ericgardner, killed, lives
#AllLivesMatter #AllLivesMatter, #PoliceLivesMatter, #WhiteLivesMatter, #NYPDLivesMatter, #BlueLivesMatter, #BlackLivesMatter, #nypd, police, people, white

Regarding the most used words, in Table 2 can be found the ten most common in each side of the protest. In both cases, the most repeated keywords are the hashtags used, while the rest defend their key ideas. It should be noted that for #AllLivesMatter, the hashtag #BlackLivesMatter appears, which indicates again, that it is not so polarized.


What sentiments did tweets express?

The sentiment in a tweet can be classified according to its polarity, that is, the emotion expressed in a sentence, in positive, neutral and negative. Additionally, it can be classified according to its subjectivity in subjective, rational and neutral. Polarity gives an idea of how people react to the events and how it makes them feel, while subjectivity measures whether this reaction is emotional or not.

Black Lives Matter

Figure 5. Percentage of tweets from #BlackLivesMatter classified as positive, neutral and negative according to their polarity, aggregated by week.

Figure 5 shows the percentage of tweets classified according to their polarity in #BlackLivesMatter over time, aggregating by week. Most of the tweets are classified as neutral, with similar values between positive and negative. However, in 2016 this trend changes with a predominant number of positive tweets, and stabilizes. This is, probably, because the most active part of the protest, which made it known worldwide, had passed.

The events related to the deaths of black people, e.g. the no indictments in Brown and Gardner cases, have the consequence of the percentage of negative tweets rising and the positive decreasing, because they express feelings of rage and injustice. However, the NBA game and the killing of two police officers have the opposite effect, as the number of positive tweets grows. In contrast, the defense of the protest by Bernie Sanders was followed by a rise of the number of neutral tweets, probably due to not being an event so emotionally charged, which encouraged other people, not involved with the protest, to use the hashtag.

Figure 6. Percentage of tweets from #BlackLivesMatter classified as rational, neutral and subjective according to their subjectivity, aggregated by week.

Regarding subjectivity, Figure 6 shows that the majority of the tweets are considered subjective, as one would expect in this type of protest. Rational tweets are around 25% while neutral are around 5% of the total. This changes in the same point as with the polarity plot, in 2016, when rational tweets constitute more than 50% of the total, while subjective are around 40%. This change, as mentioned with polarity, is likely caused by the most active and tense periods of the protest, passing.

In this case, the events related to the death of black people provoke, contrary to what could be expected, a rise in the rationality of tweets, while lowering the percentage of subjective ones. A probable reason is that these events cause more people to tweet about them, including news, whose opinion may not be so polarized. The rest of the events have the opposite effect, incrementing the number of tweets considered as subjective.

All Lives Matter

Figure 7. Percentage of tweets from #AllLivesMatter classified as positive, neutral and negative according to their polarity, aggregated by week.

With #AllLivesMatter we can observe much more variability in the classification according to polarity, shown in Figure 7. The predominant classification is still neutral, followed by positive and then negative, but in this case, there is much more variability. This is probably caused by #AllLivesMatter being more heterogeneous than #BlackLivesMatter, as was also observed with the topics of the tweets. Like before, 2016 constitutes a breaking point, with an increase of positive tweets, though in 2017 they are again surpassed by more neutral messages.

The effect of the events is not as clear as with #BlackLivesMatter. The events related to the deaths of black people tend to result in a rise of neutral tweets, while the NBA game and Bernie Sanders’ defense of the protest seem to increase the number of negative tweets. After the death of the two police officers, instead of an expected increase in the number of negative tweets, the neutral messages raised, probably because a large amount of people who did not usually employ this hashtag tweeted about it.

Figure 8. Percentage of tweets from #AllLivesMatter classified as rational, neutral and subjective according to their subjectivity, aggregated by week.

In Figure 8, we can observe that in subjectivity the same trend as with #BlackLivesMatter. Around 75% of the tweets were subjective, with 20% being rational and merely around 5% neutral until 2016, when the rational tweets started to predominate. In this case, the news about no indictment in the Brown and Gardner cases resulted in a rise of the rationality of tweets. Likewise, the NBA game and the death of the two police officers had the opposite effect, with an increment in subjective tweets. The same thing happened with the death of Sandra Bland, which gave place to much speculation.

Overall, it can be concluded in both sides we can observe similar behaviors showing that it is an emotionally charged protest, with a great number of subjective tweets, which are affected by the events.

Comparison between #BlackLivesMatter and #AllLivesMatter

Figure 9. Comparison of polarity in #BlackLivesMatter and #AllLivesMatter, aggregated by week. The tweet is negative when polarity < 0, neutral when polarity = 0 and positive when polarity > 0.

In Figure 9 we can observe a comparison of the polarity in both sides of the protest, with the data aggregated by week. Values smaller than zero indicate that the tweets were considered negative, while bigger than zero indicate positive and exactly zero neutral. The value of this measure is very similar for both #BlackLivesMatter and #AllLivesMatter. It has values close to 0, which indicates that the tweets are only slightly polarized, until 2016, when for both sides the tweets are always considered negative.

The impact of an event in polarity seems to be very similar for both sides of the protest, with a few exceptions: the events in December 2014, when the reaction was positive for #BlackLivesMatter and negative for #AllLivesMatter, and with a smaller difference with the Bernie Sanders defense of #BlackLivesMatter.

Figure 10. Comparison of subjectivity in #BlackLivesMatter and #AllLivesMatter, aggregated by week. The tweet is rational when subjectivity < 0, neutral when subjectivity = 0 and subjective when subjectivity > 0.

Figure 10 shows the subjectivity aggregated by week. With subjectivity, 0.5 indicates neutrality, while bigger values show rationality and smaller values subjectivity. Again, the values for both sides of the protest show similar values, with most of the tweets being considered rational. However, in this case they diverge in July 2016, when most tweets are considered neutral for #BlackLivesMatter, while they are still rational for #AllLivesMatter. A possible cause for this divergence is that after the most active period of the protest, the #BlackLivesMatter moved towards neutrality, while the hashtag #AllLivesMatter, as it was also employed in other protests, didn’t have this move.

Regarding specific events, it should be remarked the difference in reaction to the one-year anniversary of the death of Michael Brown in August 2015, which increased the subjective tweets with #BlackLivesMatter, while the subjectivity descended in #AllLivesMatter. The opposite happened with the Bernie Sanders’s declarations, rising subjectivity in #AllLivesMatter. It should be mentioned that even though this event defended #BlackLivesMatter, the emotional response was much bigger among the detractors of the protest than among its supporters, probably as mentioned before, due to encouraging people not involved with the protest, to tweet about it.


Graph

These graphs expose clearly the difference between the use of #BlackLivesMatter and #AllLivesMatter. #BlackLivesMatter is a hashtag that was used massively and in relation to a lot of hashtags, which in general are not connected. On the other hand, with #AllLivesMatter one can see clear communities that used each concrete hashtags.

With #BlackLivesMatter, only three different communities can be distinguished. The first one, and most important, is a general community that englobes hashtags whose only connection is the use of #BlackLivesMatter, and that includes hashtags related to this protest (#Ferguson, #ICantBreathe, #SayHerName, #PoliceBrutality, #Justice), to race (#Racism, #CivilRights, #BlackGirlMagic, #BlackCommunity, #latinos), but also to misogyny (#feminism, #WomenMarch) and politics (#Resist, #Obama, #MAGA, #ImpeachTrump). The second one is a community related to Bernie Sanders, with hashtags such as #bernie2016 or #feelthebern, which appeared due to its defense of the protest during the presidential campaign, and that also links to some mass media news (#cnn, #foxnews). Lastly, we can see a community related to Forbes, business and cars, which is due to the fact that Ford foundation is backing a Black Lives Matter fund to give exposure to the movement and that Forbes has a special section talking about this protest.

With #AllLivesMatter we can observe clearly several communities. It should be noted, that this happens even though #AllLivesMatter did not have a specific organization backing it or promoting its use. The most important community in #AllLivesMatter is the one in support of the police, characterized by the hashtags #BlueLivesMatter and #PoliceLivesMatter, which is also connected to Trump (#makeamericagreatagain, #trumptrain). In addition, there is also another community that seems to be connected to white supremacists, with the hashtags #whitelivesmatter, #whitegenocide or #teamwhite. Finally, the last community that can be observed is more general and not so centered in one concrete topic.

Other Post Formats